Abstract：Optimum techniques based on the minimum entropy principle(or variance norm)are widely used in seismic data processing. When the resolution of seismic da- ta becomes optimum in a sense，the entropy of seismic data can be thought to be minimum;that is to say,the variance norm gets to the limit value. This is so called “parsimonious criterion”.Parsimony is defined as structural ization degree which seismic record as a stochastic process exhibits in time series. Structuralization degree means a position of all possible distributions for the stochastic process that varies from disordering to ordering. To speak correctly，seismic record gets the simplest structuralization after parsimonious processing. The parsimonious processing in- cludes deconvolution，constant phase filtering，pure phase filtering，frequency spec- trum extrapolation, stacking and so on. These processes are illustrated on the mean- ing of parsimonious criterion. Some technical problems are discussed here.
